The Pitt County Study is a longitudinal investigation of anthropometric, psychosocial, and behavioral predictors of hypertension in African Americans who were aged 25-50 years at baseline in 1988. At baseline, a strong dose-response gradient was observed for alcohol consumption and blood pressure for both sexes. The current study investigated whether baseline alcohol consumption or, alternatively, changes in drinking status predicted 5-year changes in blood pressure among the 652 women and 318 men who satisfied all inclusion criteria for the longitudinal analyses. In multivariate regression analyses, baseline alcohol consumption was not significantly associated with changes in blood pressure or hypertension incidence (systolic/diastolic blood pressure > or = 160/95 mmHg) by 1993. Change in drinking status, however, was significantly associated with changes in systolic pressure. The systolic pressure increase among individuals who initiated alcohol consumption was 6.2 mmHg (95% confidence interval (CI) 1.1-6.4) greater than abstainers, while that for individuals who reported drinking at both time points was 3.8 mmHg (95% CI 1.3-11.1) greater. Blood pressure increases for persons who discontinued drinking were comparable to those of abstainers. Results were independent of baseline age, body mass index, blood pressure, and sex. Social and economic disadvantage in 1988 was significantly associated with continuation and initiation of alcohol consumption by 1993.